 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Electronics : 150V 45A MPPT - roll your own
  Page 32 of 50   |
|||||
| Author | Message | ||||
poida Guru  Joined: 02/02/2017 Location: AustraliaPosts: 1420 |
Wiseguy and Mark: I have a bag of these connectors that fit the board (again from Ali) I can supply some if needed. wronger than a phone book full of wrong phone numbers |
||||
wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
Thanks Poida and Mark, I was given an inverter yesterday "Solar River" that was certainly up the creek. It provided some donor screw terminals so thanks for your offers but I have enough for my board now. It also had the biggest most generous heatsink I think I have ever seen. I'm guessing ~ 300 x 350mm. Unfortunately transformer-less, lots of good capacitors etc but everything is glued so rigidly to the board and Im not sure I can be bothered, maybe the 2 x current sensors but I think the rest might just be binned, I'm slowly learning to be a much more selective hoarder.....  Edited 2020-09-08 01:10 by wiseguy If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
renewableMark Guru Joined: 09/12/2017 Location: AustraliaPosts: 1678 |
Yes, I found some ripper big ferrite cores in a dead GTI recently, another had big Iron choke cores similar to the large aerosharps. They are all constructed very differently, you would think they would have a cookie cutter format, but they all have their own very different ways to design them. Anyway, how do we determine to use RT1 or RT2? Cheers Caveman Mark Off grid eastern Melb |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1420 |
One of the happiest sights for me now is to see a big heatsink. A big, zero cost heatsink. OH yes. enough of that. RT1 is inductor RT2 is heatsink. Mark, maybe send me a photo of the PCB, I might see something worth the effort of keeping? I made some small changes to the latest code, to allow for temperature compensation of charge voltage. It uses the inductor NTC input as the battery temperature sensor now. I think that an inductor can easily take a bucket load of heat before causing issues so that was the one I chose to use. The menu now wants to know 3 new things for the tempco function to work. First we need an accurate temperature. This means use a 10K NTC. And we need to calibrate it using an offset value and a good thermometer (I use a K type thermocouple and a DMM) Next is the tempco itself. For my battery it's -0.120 and the unit is volts per degC. Lastly is the standard temperature for the battery voltages. Float, absorb etc. I chose 25 degC for this. How it works is the inductor NTC is measured, adjusted by the offset and saved. This temp has the "standard temperature" subtracted from it, leaving a difference which is multiplied by the tempco. For the above settings and an absorb voltage of 54V and a battery temp of 30 C we will have a corrected absorb voltage of 54 + (-0.12 * (30 - 25)) which is 53.4 V the code mpptv5_BV_tempco.zip wronger than a phone book full of wrong phone numbers |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1420 |
this is the menu now: A - max heat sink temp 65 degC 4 - H.S. throttle low 55 degC 1 - Fan ON temp 60 degC 2 - Fan OFF temp 50 degC 3 - NTC 5V or GND 0 (= 1 if 5V, = 0 if GND) C - cal. Input Volts 0.1748 ( 0.00 V) D - cal. Output Volts 0.0894 ( 0.04 V) E - cal. Input Amps 0.2401 ( -57.87 A) F - cal. Output Amps 0.2405 ( -58.38 A) G - zero Iin sensor 511 ADC counts H - zero Iout sensor 511 ADC counts I - Max battery Volts 60.00 Volts J - Absorb voltage 45.00 Volts K - Absorb tolerence 0.50 Volts 5 - Battery temp offset 22 degC 6 - BV tempco -0.20 Volts / degC 7 - Absorb BV std. temp 25.00 degC L - Maximum Current 45.00 Amps M - Night power 10.00 Watts N - Zero Energy total 0.00 kW.hr O - Equalisation voltage 15.00 Volts P - EQ duration 60.00 minutes Q - EQ reset tolerance 0.50 Volts R - Absorb duration 60.00 minutes S - Float voltage 13.00 Volts T - Float tolerence 0.50 Volts U - fast OC enable 0 (ON = 1, OFF = 0) V - pwm freq. 20kHz (20 or 40) Y - Calibrate mode 0 (1 = ON, 0 = OFF) Z - Write defaults I have a large battery temp offset of 22, since I am testing things on the bench and am using a 10K potentiometer to simulate an NTC resistor. I use an absorb voltage of 45V so that about 45 Amps is produced and sent to the 1 Ohm resistor. wronger than a phone book full of wrong phone numbers |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1420 |
It's got to 200kW.hr  No problems at all. wronger than a phone book full of wrong phone numbers |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1420 |
Now exceeded 300kW.hr This is board number 3 which I built with whatever was available. In particular, I used a small isolated 240V -> 12V supply that runs down to 40V since I used the proper 12-12 isolated supplies on previous boards. And I had no screw terminals for cables left so I soldered plain washers onto the ends of cable. A rather rough and ready assembly. This design will do the job for me. In the near future, I will make small changes to the code to support battery temperature compensation. The thermistor derived temperature is always going to be a bit out so an offset trim value will be added to the menu. And some limits to the temperature correction function.  wronger than a phone book full of wrong phone numbers |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
An easy way to measure temperature within about one degree without stuffing around with calibration would be an LM35. https://www.ti.com/lit/ds/symlink/lm35.pdf 0 to 150 Celsius = 0 to +1.5 volts output direct. Edited 2020-10-07 08:59 by Warpspeed Cheers, Tony. |
||||
| noneyabussiness Guru Joined: 31/07/2017 Location: AustraliaPosts: 513 |
A idea for terminals if interested, copper pipe hammered / crimped to cable and drill suitable hole, works connecting 2 heavy gauge wires together as well... MUCH cheaper than actual lugs, and custom size etc to boot... still can heatshink etc. Buy it at Bunnings for a decent length for 20 odd bucks.. I think it works !! |
||||
| johnmc Senior Member  Joined: 21/01/2011 Location: AustraliaPosts: 282 |
When making copper terminal lugs from flattened copper pipe, I always drill a small hole through the flattened copper, for the bolt hole, and enlarge the small hole with a tapered drift, (leaves more copper at the terminal) and stops the large drill from jamming. cheers john johnmc |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1420 |
nice parts. 0.25V at 25C is reasonable for ADC conversion. The Nano is 5V and 1024 counts so one count = 5V/1024 = 4.8mV/count or 2 counts per degree (10mV/C) This means resolution of about 0.5 deg C. Even though the LM35 is accurate to 0.1 or 0.2C the ADC will still need calibration. For a coupla years I worked in a pyrometry lab where we characterised heat treatment ovens. To get to 1 deg C uncertantiy with respect to a known and traceable standard is really good going. We sent off to NATA platinum thermocouples to be calibrated and when they returned, we used them as a standard. I like these parts, but the brainboard is not ready for 3 pin devices. It expects a resistance with 2 pins. I can happily hack the brainboard to suit the devices, change the code to use the LM35 output voltage... All I need is reasonable accuracy for a range of 15degC to 35 degC temperatures outside this range will be clamped to these extrema. I think a crappy NTC 10K will do the job with calibration. wronger than a phone book full of wrong phone numbers |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1420 |
please correct me if I'm wrong. (I am wrong. we all know that...) a 10K NTC has resistance of 10K at 25C, 15.13K at 15C, 6.75K at 35C we have the NTC on the high side of a voltage divider with a 10K resistor on the bottom side. I figure 15C means 0.33mA through the divider and this across a 10K resistor = 3.3V 35C means 6.75K, so 0.298mA through the divider so it's 2.98V across the 10K 35 - 15 C is 20 deg C range. 3.3V - 2.98V = 0.32V for 20degC range or 0.016V / degC recall the ADC is 5V for 1024 counts so that's 4.8mV/count We have 3 counts per degree. This is better than the LM35 without any amplification. wronger than a phone book full of wrong phone numbers |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
I think there is an error in your first calculation and the counts per degree calc also If you hadnt said "(I am wrong. we all know that...)" I probably would not have checked it  Hope you will still talk to me.....maybe I got it wronga 10K NTC has resistance of 10K at 25C, 15.13K at 15C, 6.75K at 35C we have the NTC as the high side resistance of a voltage divider with a 10K resistor on the bottom side. 15C means (15.13K top resistor + 10K bottom R), = 25.13K = 0.199mA through the divider chain (from 5V) and this translates to 1.99V across the 10K resistor. 35C means 6.75K top resistor + 10K bottom R, so 0.298mA through the divider so it's 2.98V across the 10K 35 - 15 C is 20 deg C range. I Get 2.98V -1.99V = 0.99V for 20degC range or 0.0495V / degC recall the ADC is 5V for 1024 counts so that's 4.8mV/count We have ~10 counts per degree (0.1Deg resolution). This is better than the LM35 without any amplification. Re calibration maybe at two points, instead of a thermistor place a 15K resistor for 15 degree cal and a 6K8 for 35 degrees cal (or for the second point another fixed resistor equivalent to say 70 degrees) job done. Edited 2020-10-07 22:58 by wiseguy If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
The bigest problem with thermistors is linearity. Also NTC thermistors really do have a negative temperature coefficient. The resistance decreases as temperature increases. That is what makes them so useful, as the resistance of most things increase with temperature. The value of the series resistor makes a HUGE difference. Try the above calculations using a 10K NTC with a series resistor of 1K, 10K, and 100K and see what happens. Thermistors are excellent, cheap, very high output, very fast response, but hopeless if you need any kind of accuracy over a fairly wide temperature range. Here are a few tricks of the trade to make them work: https://www.jameco.com/Jameco/workshop/TechTip/temperature-measurement-ntc-thermistors.html Cheers, Tony. |
||||
| SYM-1 Regular Member  Joined: 18/10/2019 Location: New ZealandPosts: 45 |
An AD590 or AD592 is 2 wire and very easy to interface and program. It outputs 1uA per degree Kelvin. A simple one point calibration and it should do the job. I used to use them with an LM336 precision op amp to get good temperature input. Edited 2020-10-08 09:48 by SYM-1 Persistence is the key |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Yup. LM334 is a very similar kind of thing. Two wire connection, measures degrees Kelvin. https://pdf1.alldatasheet.com/datasheet-pdf/view/8641/NSC/LM334Z.html Cheers, Tony. |
||||
| poida Guru Joined: 02/02/2017 Location: AustraliaPosts: 1420 |
Thanks for all your input guys. In particular, I want to thank Mike for reading and checking my work. The more you pick holes in my work the more highly I think of you. I have zero concerns due to the non-linearity of the NTC. Over the temperature range that interests me (15 - 30 C) the exponential curve fit is extraordinarily close to exact. We need this temp range for small changes to be made to the charge voltage. I will be using the 10K NTC in the design in my builds for the home and I suggest others do too. But by all means, if you want to use the 2 wire sensors or the LM35 then go ahead. Some changes to the firmware will be needed of course. I like very much the LM35 and will purchase a few to be used as calibration devices. I should expect 0.5 degC accuracy at 25 C according to the specs. I then need to measure 250mV to much better than 5mV accuracy too with the Fluke 87 which has 0.1% + 1 count accuracy on the 400mV range. 0.1% + 1 count of 250mV is 1.25mV uncertainty or 0.125 deg C due to the LM35 having a nominal 10mV /C response. I think the LM35 with the Fluke 87 will give me +/- 0.625 deg C accuracy. And please check my figures. I am wrong on a regular basis. Have a go with the NTC curve fitting if you like I choose to use the Steinhart–Hart equation from https://en.wikipedia.org/wiki/Thermistor To get the 3 coefficients visit https://sanjit.wtf/Calibrator/webCalibrator.html with the 3 above resistance and temperature combinations. I used this chart for the data. 10K NTC... https://assets.omega.com/pdf/tables_and_graphs/thermistor-resistance-kelvin.pdf I don't mind a few degrees error at 100C, that's not important, the fan and shutdown will have happened already, and if it happens 3 degrees too hot or cold that is no matter. I use https://www.altronics.com.au/p/r4112-10k-ntc-thermistor/ due to them being potted onto a metal fitting and there is not a lot of info available. Just the beta value. I think a bit of time spent with calibration will be worthwhile for my needs. wronger than a phone book full of wrong phone numbers |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Peter, one further small thought. There is also an LM34 that measures in Fahrenheit. Its still 10mV per degree, but you get a lot more millivolts to measure at any given temperature, and better use of the a/d converter range. Should then be easy to convert back to Celsius in software. https://www.ti.com/lit/ds/symlink/lm34.pdf Cheers, Tony. |
||||
| dangraham Newbie  Joined: 07/10/2020 Location: ThailandPosts: 17 |
It’s strange how coincidences go in this world. I’m sure I stumbled on this forum many years ago when working on various alternative energy installs when I lived in the U.K. but as time fly’s by by we lose track of things, and so I lost track of this forum. I had a dabble with micro-controllers many years ago but found the coding far too time consuming. In the last couple of years, I have been slowly picking up Python through some aspects at work and this has inspired me to enrolee in further python learning, and get back into using micro-controllers for my projects. I recently decided to explore DIY charge controllers, much the same as Poida, I am sick of things breaking and relying on hardware that just does not fit my needs or work as expected. I am currently based between Phuket Thailand and Fiji, with land and marine based solar systems to manage. I recently posted in another forum asking for advice on building my own solar charge controller and was directed here. I have eagerly read the 30 odd pages of this thread over the last few days and I must say this project is almost exactly aligned with my own! This is by far the friendliest and most productive internet community I have ever seen, congratulations to all of you, this is how the internet should be! My requirements are slightly different due to my battery voltage being 132v nominal, 40* 180Ah LifePo4 and the DC generator/charger tops off at 148V I will also build a 48v version as described in this thread for my electric power catamaran that is still being converted. This has 16* 280Ah LifePo4 cells and requires some serious charging power! It’s probably best to open a new thread and discuss my requirements and I am hoping you will be so kind to give me a little guidance. I’m so glad I found this forum again. |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Welcome to the Forum dangraham. Many interesting things to discuss regarding higher voltage systems, It might be better to start a new thread. This thread is now so big that a lot of interesting high voltage issues will quickly be buried and become difficult to find later on for people doing a search. Cheers, Tony. |
||||
| Page 32 of 50 |
|||||
 Print this page |
